The development of a robust process for a CRF1 receptor antagonist

Article abstract of DOI:10.1021/op100270u

A scalable and robust process was developed for the preparation of pexacerfont (2), a pyrazolotriazine corticotropin-releasing factor receptor 1 antagonist (CRF₁). The formation of the core hydroxypyrazolotriazine moiety was achieved through two consecutive cyclizations of a semicarbazide, employing reaction conditions that are significantly milder than those reported in the literature. Further conversion to the key chloropyrazolotriazine intermediate was accomplished through a novel catalytic process using phosphorous oxychloride as the chlorinating agent. The active pharmaceutical ingredient 2 was obtained in >99.5% purity with a 68% overall yield for the six synthetic steps.

Full text of DOI:10.1021/op100270u

ARTICLE
pubs.acs.org/OPRD
The Development of a Robust Process for a CRF1 Receptor Antagonist
Sꢀevrine Broxer,* Monica A. Fitzgerald, Chris Sfouggatakis, Jessica L. Defreese, Evan Barlow, Gerald L. Powers,
Michael Peddicord, Bao-Ning Su, Yue Tai-Yuen, Charles Pathirana, and James P. Sherbine
Process Research and Development, Bristol-Myers Squibb Company P.O. Box 191, New Brunswick, New Jersey 08903, United States
ABSTRACT: A scalable and robust process was developed for the preparation of pexacerfont (2), a pyrazolotriazine corticotropin-
releasing factor receptor 1 antagonist (CRF1). The formation of the core hydroxypyrazolotriazine moiety was achieved through two
consecutive cyclizations of a semicarbazide, employing reaction conditions that are significantly milder than those reported in the
literature. Further conversion to the key chloropyrazolotriazine intermediate was accomplished through a novel catalytic process
using phosphorous oxychloride as the chlorinating agent. The active pharmaceutical ingredient 2 was obtained in >99.5% purity with
a 68% overall yield for the six synthetic steps.
’ INTRODUCTION
of 7 with excess hydrazine monohydrate in refluxing toluene
afforded aminopyrazole 8 in 76% yield over the two steps from
oxazole 6. Amidine 9 was next generated via condensation of
amine 8 with ethyl acetimidate in the presence of 1 equiv of
glacial acetic acid. Subsequent treatment with diethyl carbonate
in the presence of sodium ethoxide in refluxing ethanol formed
pyrazolotriazinone 10 in 79% yield for the two steps. Chlorina-
tion was then achieved with 4 equiv of phosphorus oxychloride in
the presence of Hunig's base in toluene at reflux to generate the
key chloropyrazolotriazine intermediate 11. Finally, coupling of
intermediate 11 with (R)-2-butylamine provided the final pro-
duct 2 in ∼50% yield over the two steps.
Corticotropin-releasing factor (CRF) is a 41 amino acid peptide1
that plays a critical role in the body's response to stress.2
Hypersecretion of CRF from the hypothalamus is proposed to
trigger depression and anxiety-related disorders.3 While two
receptors have been identified, CRF binds preferentially to the
CRF1 receptor. A significant amount of research has been invested
in developing a CRF1 receptor antagonist, as studies suggest that
blocking the CRF1 receptor could treat the diseases that result
from elevated CRF levels. Accordingly, a variety of small-molecule
CRF1 receptor antagonists, containing the pyrazolo[1,5-a]-1,3,5-
triazine structural motif as in 1 (Figure 1), have been studied as
therapeutic agents for the treatment of various CRF-related
disorders.4a,4b The Discovery Chemistry group within Bristol-
Myers Squibb focused on the syntheses and structure-activity
relationships of 8-(4-methoxyaryl)pyrazolo[1,5-a]-1,3,5-triazine
CRF1 receptor antagonists as potential anxiolytic drugs. Their
efforts led to the identification of compound 2 for advancement
into clinical development as a potent CRF1 antagonist (Figure 1).5
While this sequence satisfied the initial material requirements and
was amenable to the preparation of analogs, its use for kilogram-
scale preparation of 2 presented a number of challenges. First, the
synthesis was lengthy, requiring eight isolations from the bromo-
pyridyl starting material 3 and relied on extensive use of column
chromatography for both impurity removal and isolation of non-
crystalline intermediates. A primary goal for process development
was to eliminate the use of chromatography and define process
conditions (henceforth referred to as telescope procedures) that
allowed crude noncrystalline or unstable intermediates to be directly
used in the next transformation. Telescoped processes also provide
additional efficiencies in that they minimize the number of isolation
steps. Second, the use of large quantities of highly toxic and dan-
gerously unstable hydrazine10 posed a serious challenge from a
safety perspective. As such, an alternative synthesis of aminopyrazole
8 would be required for multikilogram-scale production. Finally, the
penultimate chloropyrazolotriazine intermediate 11 was found to be
highly unstable and readily hydrolyzed when exposed to moisture.
Measures to address these stability considerations would be key
aspects for the design of the final-step chemistry and process.
A retrosynthetic analysis of 2 maintains the late-stage installation
of the chiral amine moiety, as described in the discovery synthesis,
leading to the known hydroxypyrazolotriazine intermediate 105
(Scheme 2). A disconnection strategy that relied on an alternative
cyclization at the six-position of the triazine core was then considered.
Figure 1
The synthetic strategy employed by Discovery Chemistry for
the preparation of clinical candidate 2 is outlined in Scheme 1.4-6
A key feature to the synthesis involved a Pd-mediated coupling of
6-methoxy-3-bromo-2-methyl pyridine 37 with 4-iodo-4-methyl-
isoxazole 5,8 followed by isomerization under basic conditions to
provide β-ketonitrile 7 as the sodium enolate form.9 Treatment
Received: October 7, 2010
Published: January 31, 2011
r
2011 American Chemical Society
343
dx.doi.org/10.1021/op100270u Org. Process Res. Dev. 2011, 15, 343–352
|

Organic Process Research & Development
ARTICLE
Scheme 1
In the forward sense, this transformation was envisioned to
proceed via condensation of an orthoester derivative with
aminopyrazole 12 bearing the carboxamide functionality. Further
disconnection of the pyrazole reveals semicarbazone 13 which
can be assembled by condensation of the known cyanoketone 14
and commercially available semicarbazide 15. This revised synthetic
strategy allows for a more convergent synthesis in which hydrazine
can be replaced with the less toxic semicarbazide reagent.
route utilized an alternative method to generate enolate derivate
7 based on condensation of 16 with ethyl acetate.12 This
transformation was readily achieved under basic conditions using
potassium tert-butoxide in THF to provide a mixture of cis- and
trans-enolates 17 (Scheme 3). Following neutralization and
concentration via distillation, the potassium enolate 17 was then
treated with an aqueous solution of semicarbazide hydrochloride
salt 15 in the presence of 2-propanol13 to promote crystallization
of 13 directly from the reaction media, albeit as a mixture of the
trans and semicarbazone isomers.14 The process was successfully
scaled to >50 kg to provide high-quality 13 in 96% yield as a 80:20
mixture in favor of semicarbazone 13b,15 however, a number of
challenges remained from an analytical perspective.
’ RESULTS AND DISCUSSION
Synthesis of Semicarbazone 13. The point of departure for
the modified process route from the Discovery Chemistry route
begins at the pyridyl acetonitrile 16.11 The modified process
Scheme 2
344
dx.doi.org/10.1021/op100270u |Org. Process Res. Dev. 2011, 15, 343–352

Organic Process Research & Development
ARTICLE
Scheme 3
The major challenge encountered in the preparation of 13, was
the ability to reliably analyze the isomeric product mixture. Isomers
13 exhibited poor solubility in most common organic solvents.
For the purposes of HPLC analyses, the isomers were dissolved
in DMSO/MeCN with sonication in order to facilitate full
dissolution. Once in solution, however, the isomers rapidly
interconverted, resulting in inconsistencies from assay to assay.
The sonication step also caused the formation of the cyclized
intermediate 12, as well as related impurity 8 (Scheme 4). Due
to these challenges, an indirect purity assay was developed
based on derivitization. Treatment of 13 with a solution of 0.01
wt % triethylamine (TEA) in acetonitrile resulted in full
conversion to 12 along with trace quantities of 8 as a minor
byproduct. To accurately calculate the purity, a response factor
for 12 and the relative response factor (RRF) of 12 to 8 were
measured. The chemical purity of 13 could then be estimated by
summing the area of impurity 8, weighted by the RRF, to the
area of 12.
Given that a conventional method to determine the purity
for 13 was not directly available, stringent quality criteria
were established for starting material 16 and the subsequent
penultimate hydroxypyrazolotriazine intermediate 10. This
bracketing approach to the analytical testing ensured we could
achieve the requisite purity for 10 even if the absolute purity
for 13a and 13b could not be ascertained. Furthermore,
penultimate hydroxypyrazolotriazine 10 is a robust crystalline
substance that provides excellent impurity removal. Thus, 10
was selected as a quality gate-keeper intermediate for the
synthesis.
Synthesis of Hydroxypyrazolotriazine 10. The analytical
challenges to measure the overall purity for the isomeric mixture
13 did not impact the downstream chemistry as both readily
cyclized to pyrazole 12 in the presence of 20 mol % DBU (1,8-
diazabicyclo[5.4.0]undec-7-ene) in acetonitrile at 0 °C (Scheme 5).
The subsequent addition of 25 mol % p-TsOH (p-toluenesulfo-
nic acid) and 1.3 equiv of TMOA (trimethylorthoacetate) led to
imidate 18 which smoothly cyclized to form triazine 10 within
3-5 h at reflux. The reaction workup made use of the acidic
nature of 10, since when dissolved into an aqueous sodium hydroxide
solution, the non acidic impurities selectively partitioned to the
toluene wash layer. Isolation was then accomplished by citric
acid addition to regenerate the conjugate acid, which directly
crystallized. Following filtration and drying, the product was
isolated as the monohydrate form in 66-70% yield and high
purity (99.7 AP).
While the process for 10 was suitable for use in initial scale-up
(implemented on a 10 kg input scale), there was a potential for
the product to crystallize directly from the reaction mixture. This
strategy would serve to both eliminate the cumbersome acid-base
workup and allow isolation of 10 as an anhydrous form.16,17
A
screen of reaction parameters revealed that the process could
tolerate a wide range of concentrations (5-15 L/kg) and reagent
charges (1-20 mol % DBU; 1.5 - 25 mol % p-TsOH). The
latitude of reagent stoichiometry allowed for a reduction in the
base (5 mol %) and acid (8 mol %) catalyst loadings, which
when combined with the use of toluene as the solvent during
the workup, induced the spontaneous crystallization of 10
directly from the reaction mixture. Consequently, the aqueous
workup was eliminated and high quality anhydrous hydroxy-
pyrazolotriazine 10 was isolated in 75-80% yield on laboratory
scale.
Scheme 4
Implementation of this process on a 55 kg scale in the pilot
plant afforded hydroxypyrazolotriazine 10 in 71% isolated yield,
which was comparable to laboratory scale. However, the amount
of 10 remaining in the mother liquors was greater than expected
(16% yield loss). Further examination of reaction conditions
indicated that the solubility of 10 in toluene was dramatically
increased by increasing DBU levels.18 Accordingly, a simple way
to optimize the recovery of 10 was to reduce the amount of DBU
further to 1 mol %.
Understanding the reaction mechanism and decomposition
pathways further aided the process development and optimization.
The first reaction consists of isomers 13a and 13b cyclizing to
form a single intermediate 12 within 1 h at ambient temperature.
Extended reaction times resulted in partial urea cleavage to
345
dx.doi.org/10.1021/op100270u |Org. Process Res. Dev. 2011, 15, 343–352

Organic Process Research & Development
ARTICLE
Scheme 5
pyrazole 8 (Scheme 4). This was solved by a reduction in
operating temperature to <5 °C, which controlled impurity 8
to acceptable levels (<0.5%) within a 2 h reaction time frame.
Following the addition of p-TsOH and TMOA, the corre-
sponding imidate 18 (Scheme 5) was observed as the major
species within 30 min at reflux. Heating the reaction mixture at
reflux for an additional 3 h resulted in complete conversion to
cyclized product 10.
This impurity formed via two distinct pathways. In the first, the
amide group is initially cleaved to afford 8 and subsequent
TMOA/TFA addition formed 21 (pathway a, Scheme 8). The
Scheme 6
During the conversion of 12 to 10, approximately 10% of a
transient species was observed within 1-2 h, but had entirely
disappeared after 3 h. LCMS analysis revealed that the unknown
compound had the same mass as imidate 18, suggesting the
possibility of an acyl group migration. To confirm this hypothesis, an
isolated sample of pyrazole 12 was heated to 100 °C in toluene,
resulting in a 90:10 mixture of 12 and the unknown compound
which was subsequently identified as acyl migration product 19
by NMR analyses (Scheme 6). Similarly, heating a sample of
isolated isomer 19 produced the same equilibrium 90:10 mixture.
We surmised that under the reaction conditions (p-TsOH/
TMOA), isomer 19 generated the corresponding imidate 20
(Scheme 7), as suggested by LCMS analysis. By analogy, one
would expect a similar equilibrium to exist between 20 and 18
as between 12 and 19. As intermediate 18 can efficiently close
to afford 10, imidate 20 remains a productive intermediate
for the formation of 10.
The use of p-TsOH as catalyst was not optimal, owing to its
reaction with the methanol generated during the conversion of
12 to 10 to produce the corresponding genotoxic p-toluene
sulfonate methyl ester.19 Although the p-TsOH methyl ester was
formed in low levels in 10 (<150 ppm), it was effectively removed
during crystallization and could not be detected in the final
product 2. Nevertheless, we preferred to use an alternative acid to
avoid this concern. Several acids (HCO2H, AcOH, HCl, TFA)
were screened to determine if they were suitable replacements
for p-TsOH. TFA was selected as the replacement because it
provided an improved reaction profile with respect to acyl
migration and des-acyl impurities 19 and 8, and its correspond-
ing methyl ester derivative is a nongenotoxic impurity.
formation of 8 could be minimized by charging the DBU
at low temperature (<5 °C) under anhydrous conditions
(KF <600 ppm). Alternatively, cleavage of intermediate 18
(pathway b) could also form 21. This pathway was avoided by
using a substoichiometric amount of TFA (20 mol %) and
charging TMOA at 0 °C.
Several large-scale laboratory reactions (100-650 g) were
performed to test the optimized reaction conditions described
above. The product losses to the mother liquor were reduced to
4% with a corresponding higher isolated yield of 86% (vs 71%)
and a purity of 99.7 AP.
While the first-generation process addressed concerns regard-
ing the trans and semicarbazone isomers (13a and 13b), and
provided high-quality penultimate hyroxypyrazolotriazine 10, a
more efficient and robust process was required for further scale-up.
This goal was realized through the course of development work
by understanding the reaction pathway, impurity formation, and
material losses. By eliminating the acid/base workup, reducing
reagent charges, and identifying TFA as a replacement for p-TsOH,
an efficient multikilogram-scale process for the preparation of
anhydrous 10 was developed. More importantly, the yield and
cycle time improvements did not compromise quality, and the
high level of purity was maintained while eliminating the potential
Although the performance of TFA as a catalyst was superior to
p-TsOH, the issue of the overall low yield remained. This was
attributed, in part, to an unproductive side-product, imidate 21,
which forms upon cleavage of the amide moiety (Scheme 8).
346
dx.doi.org/10.1021/op100270u |Org. Process Res. Dev. 2011, 15, 343–352

Organic Process Research & Development
ARTICLE
Scheme 7
Scheme 8
for a genotoxic impurity (GTI) as demonstrated from the process
comparisons shown in Table 1.
Optimizing the API Step: Evolution of the Base-Catalyzed
Chlorination. The final transformation in the synthesis was to
append the (R)-2-butylamine side chain. This is most efficiently
accomplished by electrophilic activation of the core via conversion
of the hydroxyl group to the chloride, followed by nucleophilic
displacement with the amine. Chlorinations of hydroxypyridine
systems such as compound 10 are well established in the
literature and are often performed using neat refluxing POCl3.20
However, reactions using these conditions present a significant
scale-up challenge from both a safety and environmental standpoint.
We sought to accomplish the chlorination of compound 10 under
milder conditions.
Our initial approach to 2 through chlorination of hydroxypyr-
azolotriazine 10 involved heating a solution of 10 in toluene with
diisopropylethylamine (DIPEA), POCl3 and benzyltributylam-
monium chloride (BTAC)21 at 80 °C.22 The BTAC provides a
chloride ion source which has been shown to increase the rate
of reaction.23 Once the chlorination was complete, 1.7 equiv of
(R)-2-butylamine and 1 equiv of DIPEA were added to furnish 2
in 42-57% yield (entry 1, Table 2).
The process posed several challenges during the initial scale-
up investigations. The presence of the stoichiometric amount
of BTAC resulted in a biphasic reaction mixture that made
scalability, mixing, and sampling unreliable. The reaction workup
in this case furnished a red, oily solution that required filtration
through silica gel for color removal. Additionally, as 2 is very
soluble in toluene (370 mg/mL), a lengthy solvent exchange to
octane was required to achieve the desired solvent composition
for crystallization.
Table 1. Hydroxypyrazolotriazine (10) process comparison
first
second
third
generation
generation
generation
acid cat.
base cat.
p-TsOH
p-TsOH
(0.08 equiv)
DBU
(0.05 equiv)
TFA
(0.2 equiv)
DBU
(0.01 equiv)
(0.2 equiv)
DBU
(0.2 equiv)
extractions
titrations
3
0
0
1
0
0
1
2
distillations
isolated form
yield (%)
1
monohydrate
anhydrous
71
anhydrous
68
86
HPLC (AP)
potential GTI
99.7
yes
99.7
99.7
no
yes
347
dx.doi.org/10.1021/op100270u |Org. Process Res. Dev. 2011, 15, 343–352

Organic Process Research & Development
ARTICLE
Table 2. Reaction conditions examined for chlorination of 10a
entry
solvent
toluene
additive
temperature (°C)
time (h)b
yield (%)c
1
2
3
4
5
0.1 equiv BTAC
0.6 equiv BTAC
none
70-75
70-75
70-75
20
2-4
2-4
1
42-57
59-70
62
toluene/MeCN
MeCN
MeCN
0.01 equiv DABCO
none
1
78
90d
MeCN
20
48
a 1.3 equiv of POCl3, 2.0 equiv of DIPEA; 1.0 equiv of DIPEA, 1.7 equiv of (R)-2-butylamine. b Determined as >98% conversion to 11 by HPLC analysis.
c Based on isolated 2. d Conversion to 11.
From a quality perspective, the presence of a dimeric impurity,
formed during the course of the chlorination, proved challenging
as well. On the basis of LC/MS data and literature precedent,24
we anticipated the formation of the symmetric dimer 22 (Figure 2),
however, single-crystal X-ray analysis revealed it was the regioi-
someric dimer 23.25 This dimeric impurity was observed at levels
greater than 1% based on HPLC area, and could not be purged to
acceptable levels by recrystallization of 2.
Although the initial strategy was to qualify 23 to a level of 1.0
AP in toxicological studies, we continued to study methods to
further lower its level and streamline the process. Reducing the
amount of BTAC to 0.6 equiv and adding 0.6 volumes of MeCN
resulted in a homogeneous solution that allowed for representative
in-process sampling and provided a yield improvement of approxi-
mately 15% (entry 2, Table 2). Subsequently, it was found that
BTAC was not critical for the chlorination step if performed in
MeCN as solvent (entry 3, Table 2).26 Increasing the volume of
MeCN compared to the toluene process (15 vs 8 L/kg) also
reduced the amount of 23 by 50%. In addition, the solubility
profile of 2 in MeCN allowed for isolation of the final product
directly from the reaction mixture by crystallization from
MeCN/H2O.
Since this impurity is a potential alkylating agent, its level required
close monitoring in the final product. We hypothesized that the
formation of 24 was the result of nucleophilic attack of DABCO
on the proposed O-phosphorylated intermediate 25. Presum-
ably, chloride ion displacement of DABCO (pathway a) provides
the desired product 11. Conversely, attack at either of the three
R-carbons of the activated DABCO intermediate 26 (pathway b)
would produce impurity 24.
The dealkylation of tertiary amines during chlorinations of
compounds such as 10 with POCl3 is known,29 as was the opening
of DABCO ammonium salts by chloride.30 As such, we focused
our attention toward identification of an alternate catalyst that
would mitigate the formation of an impurity that has the
potential to act as an alkylating agent. A brief summary of the
tertiary amines surveyed is shown in Table 3. The acyclic and less
sterically strained tertiary amines such as triethyl- and triphenyl-
amine (entries 1 and 2), as well as pyridine and N-methylpyrro-
lidine (entries 3 and 4), were less active than DABCO in catalyzing
the chlorination. Comparable conversions to 11 were observed
Earlier work on previous CRF drug candidates revealed that
DABCO (1,4-diazabicyclo[2.2.2]octane) could be used as a catalyst
to accelerate the chlorination reaction.27 Toward this end, we
were pleased to find that the presence of 1 mol % of DABCO
afforded complete conversion to chloropyrazolotriazine 11 with-
in 1 h at 20 °C (entry 4, Table 2), and further reduced byproduct
23 to 0.2 AP. By comparison, the reaction in MeCN in the
absence of DABCO required 48 h to reach 90% conversion (entry 5,
Table 2). This highlights the powerful catalytic properties of
DABCO for the chlorination.28
To enable process optimization, additional impurities were
characterized in an effort to gain a deeper understanding of the
reaction profile and to improve the overall quality of the final
product. The DABCO-catalyzed chlorination gave rise to e0.5%
of a new impurity identified as the chloro-derivative 24 (Scheme 9).
Figure 2
348
dx.doi.org/10.1021/op100270u |Org. Process Res. Dev. 2011, 15, 343–352

Organic Process Research & Development
ARTICLE
Scheme 9
The reaction promoted with 6 mol % N-methylmorpholine
(NMM) resulted in complete conversion to 11 within 12 min
(entry 7, Table 3), whereas other derivatives of morpholine were
less efficient (entries 8 and 9). While other saturated six-membered
heterocyclic amines were effective catalysts (entries 10 and 11),
we selected NMM due to its higher catalytic activity .
The stoichiometry of NMM was optimized at 0.5 mol % which
led to >98% conversion to 11 within 1 h at 25 °C, although the
dealkylated impurity 27 (Figure 3) still formed in analogy to the
DABCO catalyzed reaction. The decrease in catalyst loading
reduced the formation of the dealkylated impurity 27, which forms
at levels proportional to the amount of catalyst used. Importantly,
a chlorinated impurity analogous to 24 was not observed in the
NMM-catalyzed process since any undesired chloride ion attack
would presumably occur on the primary methyl group of NMM
to form trace amounts of chloromethane gas.
In addition to the dimer and catalyst-related impurities, four
impurities related to DIPEA incorporation were identified (28-
31, Figure 4). Dealkylated DIPEA impurities 28 and 29, ob-
served during the chlorination step, are thought to form either by
elimination or by attack of chloride ion on either the isopropyl or
ethyl groups of a DIPEA-activated intermediate analogous to 26
(Scheme 9). The formation of both impurities was minimized
31
below 0.15% by a rapid addition of POCl3.
Following the chlorination, compound 11 was telescoped
into the amine displacement reaction via in situ treatment with
(R)-2-butylamine and an additional stoichiometric equivalent
of DIPEA to effect the transformation to the final product 2.
The displacement of the chloride occurs readily at -5 °C, but is
also accompanied by the formation of two additional DIPEA-
related impurities (30-31, Figure 4) that were observed on
scale. Enamine impurity 30 and C-bound adduct 31 were
envisioned to form through the iminium salt of DIPEA.32,33
The enamine tautomer can displace the chloride ion in 11 to
form impurity 30. Alternatively, the initial iminium ion could be
trapped by nucleophilic addition of 11 to form 31. Although
levels of 30 were as high as 1% during the reaction, it was
completely purged during the isolation. Unfortunately, impurity
31 was formed in variable amounts (0.1 - 0.5%) during the
reaction and did not purge during the crystallization. While rapid
addition of (R)-2-butylamine minimizes its formation, such a
practice was not feasible on scale given the exothermic nature of
the addition (Tad = 64 °C). Reversing the order of amine additions
((R)-2-butylamine followed by DIPEA) effectively controlled
the formation of 31 to below detection. All of these impurities
were effectively controlled by either the order or rate of addition
of reagents and by the workup conditions.
when amines that were structurally similar to DABCO were used
(entries 5 and 6); however, in order to avoid chlorinated impurities,
these amines were not pursued.
Table 3. Examples of tertiary amine catalysts used in chlori-
nation of 10
Figure 3
349
dx.doi.org/10.1021/op100270u |Org. Process Res. Dev. 2011, 15, 343–352

Organic Process Research & Development
ARTICLE
Figure 4. DIPEA-related impurities.
Early iterations of the process required multiple crystallizations of
the final product to meet the required purity specifications,
whereas the process improvements described above dramatically
improved the product quality and yield, without the need for a
recrystallization (Table 4).
Furthermore, the optimizations to the reaction portion of the
process allowed for a simplification of the reaction workup in the
telescoped preparation of 2. The final optimized process used
aqueous NaOH to quench the excess phosphorous chloride,
followed by heating to 65 °C to dissolve the resulting slurry.
The solution was then cooled and diluted with water to induce
crystallization, and the product was isolated by filtration. The two-
step chlorination/amination sequence, provided high-quality drug
substance 2 in a 82% yield (>99.8 AP, 99.2 wt %, and 99.8% chiral
purity) on greater than 40 kg scale.
in six steps (three isolations) with an overall yield of 68%
(Scheme 10).
’ EXPERIMENTAL SECTION
The experimental outlined below provide representative pro-
cedures for what was run on scale in the kilo/pilot plant facilities.
Reagents and solvents were acquired from a variety of sources
1
and employed without further purification. H NMR spectra
were obtained at 400 MHz, 13C NMR spectra were obtained at
100 MHz at 25 °C in CDCl3 except as indicated. Chemical shifts
are reported in ppm downfield from an internal tetramethylsilane
standard or relative to the residual solvent signal. Coupling constants
(J) are given in hertz. Moisture determinations were performed
with a Karl Fischer titrator. HPLC analyses were performed using
a reverse phase technique. Infrared spectra were recorded in an
FT instrument at 4.00 cm-1 resolution.
Table 4. API (2) Process Comparison
2-(4-Methoxy-2-methyl-3-pyridyl)-3-methylbut-2-enylnitrile
Semicarbazide (13a) and 2-(4-Methoxy-2-methyl-3-pyridyl)-
3-oxobutyronitrile Semicarbazone (13b). To an anhydrous
solution of ethyl acetate (169 L) and tetrahydrofuran (242 L)
(KF <600 ppm), was added 16 (69.9 kg, 432 mol), followed by a
solution of potassium tert-butoxide (303 L, 496 mol, 1.0 M in
tetrahydrofuran). The reaction mixture was heated to 45 °C, and
HPLC analysis indicated >98% conversion to enolate 17 within
30 min. The addition of H2O (105 L) was followed by vacuum
distillation at an average rate of 260 L/h. The jacket temperature
was gradually increased to 65 °C, and the reaction mixture
temperature remained below 25 °C during the vacuum distillation.
Once ∼400 L remained, the jacket temperature was decreased to
25 °C, and isopropyl alcohol (140 L) was charged followed by a
solution of semicarbazide hydrochloride salt (96.3 kg, 863 mol) in
water (385 L). The reaction was held for 4 h at 35 °C at which time
HPLC analysis indicated >99% conversion to the isomeric mixture
13. The product crystallized as the reaction progressed. The tem-
perature was cooled to 5 °C and held for 7 h. The reaction mixture
was filtered on a filter dryer and washed with water (3 ꢀ 420 L) to a
conductivity specification of <3 mS/cm. The product was dried with
an initial nitrogen sweep for about 90 min and then at 50 °C/50
mbar for 18 h. Product 13 was obtained as an off-white solid, 109 kg
(96% corrected yield, 95.8 wt %, 100 area % at 220 nm).
BTAC
DABCO
Process
NMM (final)
process
process
chlorination temp
extractions
80 °C
20-25 °C
20-25 °C
0
3
3
distillations
1
1
0
polish filtration
final solvent
yes (silica gel)
toluene/
octane
63%
no
no
acetonitrile/
water
78%
acetonitrile/
water
82%
system
Yield (crude)
HPLC AP (crude)
recrystallization
yield (recrys)
overall yield
>98.85
yes
>97.93
yes
>99.8
no
81%
81%
N/A
82%
53%
61%
’ CONCLUSION
A robust and scalable process was developed for the synthesis
of 2. The key challenges addressed were an understanding of
the complex reaction sequence in the synthesis of hydroxypyr-
azolotriazine 10, the development of a high-yielding and more
environmentally friendly process for the chlorination step,
and the characterization and origin of impurities found in the
final drug substance. A significant improvement in cycle time,
yield, and quality were achieved at each stage of development,
resulting in a process that can be safely and reproducibly run
on scale. With the improvements described, an efficient and
commercially viable synthesis of CRF candidate 2 was achieved
1
trans-13a: H NMR (400 MHz, DMSO-d6) δ 2.09 (s, 3H),
2.35 (s, 3H), 3.34 (s, 3H), 5.93 (broad s, 2H), 6.64 (d, J = 12.0
Hz, 1H), 7.52 (d, J = 12.0 Hz, 1H), 7.82-7.86 (broad m, 2H);
13C NMR (100 MHz, DMSO-d6) δ 17.0, 22.6, 53.3, 74.3, 108.5,
120.1, 122.2, 142.4, 155.5, 159.3, 162.6.
350
dx.doi.org/10.1021/op100270u |Org. Process Res. Dev. 2011, 15, 343–352

Organic Process Research & Development
ARTICLE
Scheme 10. Optimized synthesis of 2a
a Reagents and conditions: (i) 1.15 equiv of t-BuOK in THF, 4.0 equiv of EtOAc, THF; (ii) 2.0 equiv of 15, H2O/i-PrOH (96%); (iii) 1 mol % DBU,
MeCN; (iv) 20 mol % TFA, 1.3 equiv of H3CC(OMe)3, MeCN (86%); (v) 0.5 mol % N-methylmorpholine, 1.3 equiv of POCl3, 2.0 equiv of DIPEA,
MeCN; (vi) 1.7 equiv of (R)-2-butylamine, 1.0 equiv of DIPEA, MeCN (82%)
1
Semicarbazone-13b: H NMR (400 MHz, DMSO-d6) δ
1.72 (s, 3H), 2.44 (s, 3H), 3.84 (s, 3H), 5.37 (s, 1H), 6.74 (d, J =
12.0 Hz, 1H), 7.61 (d, J = 12.0 Hz, 1H), 9.54 (s, 1H); 13C NMR
(100 MHz, DMSO- d6) δ 15.5, 23.0, 42.2, 54.2, 109.0, 118.4,
119.9, 140.0, 141.7, 154.7, 157.0, 162.7.
160.9; HRMS Calcd for C14H16N5O2 286.1304 [M þ H]; found
286.1313 [M þ H].
4-(2-R-Butylamino)-2,7-dimethyl-8-(2-methyl-6-methoxy-
3-pyridyl)[1,5-r]-pyrazolo-1,3,5-triazine (2). To a slurry of 10
(42.2 kg, 148 mol) in dry (<1000 ppm) acetonitrile (232 L) was
added N-methylmorpholine (83 mL, 740 mmol), and N,N-
diisopropylethylamine (51.5 L, 296 mol). After cooling to 0 °C,
phosphorous oxychloride (17.9 L, 192 mol) was added, then the
temperature was increased to 25 °C and held until HPLC analysis
indicated >98% conversion (1 h). Upon completion of the chlori-
nation reaction, the batch was cooled to -5 °C and MeCN (32 L)
was charged. (R)-2-Butylamine (25.4 L, 251 mol) was charged
over 30 min to maintain the batch temperature <30 °C, followed
by N,N-diisopropylethylamine (25.8 L, 148 mol). The tempera-
ture was increased to 25 °C and after 1 h, HPLC analysis indicated
>98% conversion to 2. The reaction was quenched with water
(54.3 L) and 3 N sodium hydroxide (109 L, 325 mol) to a pH <7.
The slurry was heated to 65 °C to provide a homogeneous solution,
and the reaction mixture filtered through a 10 μm polish filter,
followed by an acetonitrile rinse (32 L). The temperature was
lowered to 55 °C and water (42 L) was added until a slurry was
observed (1 h). Subsequently, the temperature of the batch was
lowered to 23 °C, and additional water (295 L) was charged
to complete crystallization of the batch. After a 1-h hold, the
slurry was transferred to a Hastelloy filter drier with a 10 μm
polypropylene cloth, then was washed with 1:1 water/acetonitrile
(84 L), followed by water (6 ꢀ 170 L) until the conductivity was
<30 μS/cm. The product was dried on the filter at 50 °C and
50 mmHg for 16 h, and the product was obtained as a white solid
(41.2 kg, 82% corrected yield, 99.2 wt %, 99.8% chiral purity,
99.8 area % at 220 nm).
HRMS Calcd for C12H15N5O2 261.1226; found 261.1229; Anal.
Calcd for C12H15N5O2: C, 55.16; H, 5.78; N, 26.80. Found: C,
55.03; H, 5.81; N, 26.99.
8-(6-Methoxy-2-methyl-3-pyridyl)-2,7-dimethylpyrazolo-
[1,5-r][1,3,5]triazin-4-ol (10). A solution of 13 (650 g, 2.5 mol)
in acetonitrile (6.54 L) at 5 °C was treated with a solution of 1,8-
diazabicyclo[5.4.0]undec-7-ene (3.75 mL, 24.9 mmol) in acet-
onitrile (109 mL). The reaction was held for 6 h at which time
HPLC analysis indicated >98% conversion to intermediate 12.
Trifluoroacetic acid (37.7 g, 495 mmol) was added followed by
the addition of trimethyl orthoacetate (413 mL, 3.24 mol) and a
rinse with acetonitrile (435 mL). The jacket temperature was
increased to 90 °C over 1 h and held 6 h at which time HPLC
analysis indicated >98% conversion to 10. The temperature was
lowered to 60 °C and seeded with 10 (3.16 g, 11.1 mmol). After
holding at 60 °C for 1 h, the temperature was lowered over 4 h to
20 °C. A solvent exchange to toluene was accomplished by first
removing acetonitrile by distillation at 35 °C and 150 Torr to a
volume of 3.25 L. Subsequently, toluene (6.5 L, 62 mol) was charged,
and a distillation at 55 °C and 150 Torr served to remove additional
acetonitrile. The batch was cooled to 0 °C and aged overnight. The
slurry was filtered, washed with toluene (3 ꢀ 1.3 L), and dried under
vacuum at 50 °C. Product 10 was obtained as a white solid, 616 g
(86% corrected yield, 99.9 wt %, 99.9 area % at 220 nm).
1H NMR (400 MHz, CDCl3) δ 2.28 (s, 3H), 2.34 (s, 3H),
2.48 (s, 3H), 3.95 (s, 3H), 6.67 (d, J = 12.0 Hz, 1H), 7.43 (d, J =
12.0 Hz, 1H); 13C NMR (100 MHz, CDCl3) δ 11.2, 19.2, 20.3,
51.7, 105.3, 108.4, 115.5, 139.9, 143.3, 143.8, 151.6, 153.2, 153.8,
1H NMR (400 MHz, CDCl3) δ 0.79 (t, J = 12.0 Hz, 3H), 1.11
(d, J = 12.0 Hz, 3H), 1.47 (m, 2H), 2.06 (s, 3H), 2.12 (s, 3H),
351
dx.doi.org/10.1021/op100270u |Org. Process Res. Dev. 2011, 15, 343–352

Organic Process Research & Development
ARTICLE
2.25 (s, 3H), 3.72 (s, 3H), 4.04-4.13 (m, 1H), 6.01 (broad d, J =
12.0 Hz, 1H), 6.41 (d, J = 12.0 Hz, 1H), and 7.19 (d, J = 12.0 Hz,
1H); 13C NMR (100 MHz, CDCl3) δ 10.3, 13.1, 20.4, 22.8, 26.0,
29.6, 48.0, 53.2, 106.5, 107.4, 118.4, 141.5, 146.5, 147.6, 153.6,
155.6, 162.8, 163.7; HRMS Calcd for C18H25N6O 341.2090
[M þ H]; found 341.2088 [M þ H]; Anal. Calcd for C18H24-
N6O: C, 63.5; H, 7.10; N, 24.68. Found: C, 63.49; H, 7.04;
N, 24.98.
J. Am. Chem. Soc. 1951, 73, 2239–2241. (h) Page, G. A.; Tarbell, D. S.
J. Am. Chem. Soc. 1953, 75, 2053–2055.
(12) To ensure full dissolution of 16, 4 equiv were necessary.
(13) Reaction of enolate 17 with semicarbazide 15 in THF/H2O as
solvent did not lead to appreciable levels of conversion.
(14) Pure regioisomers trans-13a and semicarbazone 13b were
isolated and used for characterization. The cis-isomer was not observed.
(15) The isomer ratio was estimated by careful solid-state NMR
techniques as 13a and 13b were observed to readily interconvert in
solution.
’ AUTHOR INFORMATION
(16) Isolation of 10 as a monohydrate required an azeotropic
distillation from MeCN to remove H2O prior to the start of the
subsequent API step.
(17) Heating anhydrous 10 to temperatures above 50 °C in the
presence of H2O afforded the monohydrate. Prolonged exposure of
anhydrous 10 as a slurry in the presence of water at room temperature
did not produce the monohydrate form.
Corresponding Author
*E-mail: sevrine.broxer@bms.com
’ ACKNOWLEDGMENT
(18) With 1 mol % of DBU, the solubility of 10 is 2 mg/mL whereas
with 5 mol % it is 25 mg/mL.
We thank our engineering colleagues Otute Akiti, Martina
Olzog, Nathan Domagalski, Victor Hung, Brent Nielsen, and
Richard Schild for enabling implementation into the pilot plant.
We also thank Christopher Wood, John Castoro, Ruben Lozano,
Mariann Neverovitch, and Michael Galella for analytical support,
and Fucheng Qu for providing the initial observations on
the DABCO-catalyzed chlorination. Finally, we acknowledge
Lindsay A. Hobson and Rodney L. Parsons, Jr., for helpful
discussions, together with PR&D senior management for support
during the preparation of the manuscript.
(19) Snodin, D. J. Regul. Toxicol. Pharmacol. 2006, 45, 79–90.
(20) Anderson, N. G.; Ary, T. D.; Berg, J. L.; Bernot, P. J.; Chan,
Y. Y.; Chen, C.-K; Davies, M. L.; DiMarco, J. D.; Dennis, R. D.;
Deshpande, R. P.; Do, H. D.; Droghini, R.; Early, W. A.; Gougoutas,
J. Z.; Grosso, J. A.; Harris, J. C.; Haas, O. W.; Jass, P. A.; Kim, D. H.;
Kodersha, G. A.; Kotnis, A. S.; LaJeunesse, J.; Lust, D. A.; Madding,
G. D.; Modi, S. P.; Moniot, J. L.; Nguyen, A.; Palaniswamy, V.;
Phillipson, D. W.; Simpson, J. H.; Thoraval, D.; Thurston, D. A.; Tse,
K.; Polomski, R. E.; Wedding, D. L.; Winter, W. J. Org. Process Res. Dev.
1997, 1, 300–310.
(21) Robins, M.; Uznanski, B. Can. J. Chem. 1981, 59, 2601–2606.
(22) Similar conditions were utilized for a related molecule, see ref 6.
(23) The addition of HCl or LiCl is thought to collapse the
intermediate dichlorophosphinate ester, see: (a) Fujino, K.; Takami,
H.; Atsumi, T.; Ogasa, T.; Mohri, S.-i; Kasai, M. Org. Process Res. Dev.
2001, 5, 426–433. (b) Connolly, T. J.; Matchett, M.; Sarma, K. Org.
Process Res. Dev. 2005, 9, 80–87.
’ REFERENCES
(1) Vale, W.; Spiess, J.; Rivier, C. Science 1981, 213, 1394–1397.
(2) For recent reviews, see: (a) Gilligan, P. J. Expert Opin. Ther.
Patents. 2006, 16, 913–924. (b) Arzt, E.; Holsboer, F. Trends Pharmacol.
Sci. 2006, 27, 531–538. (c) Berton, O.; Nestler, E. J. Neuroscience 2006,
7, 137–151.
(3) Holsboer, F. Neuropsychopharmacology 2000, 23, 477–501.
(4) (a) Gilligan, P. J.; Clarke, T.; He, L.; Lelas, S.; Li, Y.-W.; Heman,
K.; Fitzgerald, L.; Miller, K.; Zhang, G.; Marshall, A.; Krause, C.;
McElroy, J. F.; Ward, K.; Zeller, K.; Wong, H.; Bai, S.; Saye, J.; Grossman,
S.; Zaczek, R.; Arneric, S. P.; Hartig, P.; Robertson, D.; Trainor, G.
J. Med. Chem. 2009, 52, 3804–3092. (b) Gilligan, P. J.; He, L.; Clarke, T.;
Tivitmahaisoon, P.; Lelas, S.; Li, Y.-W.; Heman, K.; Fitzgerald, L.; Miller,
K.; Zhang, G.; Marshall, A.; Krause, C.; McElroy, J.; Ward, K.; Shen, H.;
Wong, H.; Grossman, S.; Nemeth, G.; Zaczek, R.; Arneric, S. P.; Hartig,
P.; Robertson, D. W.; Trainor, G. J. Med. Chem. 2009, 52, 3073–3083
and references cited therein.
(24) Fujino, K.; Takami, H.; Atsumi, T.; Ogasa, T.; Mohri, S.-i.;
Kasai, M. Org. Process Res. Dev. 2001, 5, 426–433.
(25) A similar dimeric impurity has been observed, see ref 20.
(26) The influence of solvent for the chlorination reaction is under
investigation and will be reported in due course.
(27) Recently, patents reporting the use of DABCO as catalyst were
published, see: (a) Aiqing F. Zhuanli F. Shuomingshu G. S., CN
101423498 A 20090506 (CAN 150:563847 AN 2009:561663), 2009.
(b) Aiqing F. Zhuanli F. Shuomingshu G. S. CN 101445485 A 20090603
(CAN 151:78133 AN 2009:689505), 2009.
(28) The exact role of DABCO in the catalyzed chlorination is
currently under investigation and will be disclosed in due course.
(29) (a) Robins, R.; Christensen, B. E. J. Am. Chem. Soc. 1952, 74,
3624–3627. (b) Robins, M.; Uznanski, B. Can. J. Chem. 1981, 59, 2601–
2607. (c) Boyle, P. H.; Gillespie, P. J. Chem. Res. (S) 1989, 9, 282 also see
ref 23.
(5) Gilligan, P. J. PCT Int. Appl. WO/2002/072202, 2002.
(6) He, L.; Gilligan, P. J.; Zaczek, R.; Fitzgerald, L. W.; McElroy, J.;
Shen, H-S.L.; Saye, J. A.; Kalin, N. H.; Shelton, S.; Christ, D.; Trainor, G.;
Hartig, P. J. Med. Chem. 2000, 43, 449–456.
(7) Gray, M. A.; Konopski, L.; Langlois, Y. Synth. Commun. 1994, 24,
1367–1379.
(30) Doronina, S. O.; Gall, A. A.; Shishkin, G. V. Khim. Geterotsikl.
Soedin. 1992, 8, 1091–1094.
(8) Kochetkov, N. K.; Sokolov, S. D.; Vagurtova, N. M. Zh. Obshch.
Khim. 1961, 31, 2326–2333.
(31) On 40 kg input scale, 30 kg of POCl3 was added in 4 min,
resulting in an increase of batch temperature from -4 to 3.5 °C with the
jacket temperature set to -10 °C.
(32) Schreiber, S. Tetrahedron Lett. 1980, 21, 1027–1030.
(33) Enamine formation may occur via hydride transfer as described
in ref 32 or as shown below:
(9) Zhou, J.; Oh, L. M.; Ma, P. U.S. Pat. 6,562,965, 2003.
(10) Hydrazine Hazard Summary; United States Environmental
Protection Agency, (created in April 1992; revised in January 2000;
updated in November 2007) 2007; http://www.epa.gov/ttn/atw/
hlthef/hydrazin.html (accessed September 9, 2010).
(11) The pyridyl starting material was obtained from commercial
sources or by custom synthesis, see: (a) Miyata, H.; Oki, K.; Morikawa,
K. JP 2001302658, 2001. (b) Yu, Z. CN 1088574, 1994;(c) Nishida, A.;
Shirakawa, S.; Harada, T.; Fujisaki, S.; Kajigaeshi, S. Technol. Rep.
Yamaguchi Univ. 1988, 4, 145–50. (d) Nagel, D. L.; Kupper, R.;
Antonson, K.; Wallcave., L . J. Org. Chem. 1977, 42, 3626–3628. (e)
Stogryn, E. L. J. Org. Chem. 1972, 37, 673. (f) Meyers, A. I.; Collington,
E. W. J. Org. Chem. 1971, 36, 3044–3045. (g) Rapoport, H.; Campion, J.
352
dx.doi.org/10.1021/op100270u |Org. Process Res. Dev. 2011, 15, 343–352